Bestsellers vs. Personalized Recommendations – Who wins?
I can’t count how many times I’ve been asked by a client to use top products when we were implementing new campaigns on popups, 404 pages, 0 search results, etc. But it’s easy to count how many times I’ve recommended them, to use personalised recommenndation instead – it happens in 100% of cases. Why? First, I intuitively feel it will performs better, but if you don’t trust my intuition (I can’t blame you), I have proof that personalized offers simply work better than top products.
Top products include a range of metrics which allow you to find bestsellers, most-visited products and items which generate the most revenue. And it’s tempting to use them - they are popular, so it seems a safe bet that they will interest users. But do they really? Isn’t it better to focus on particular customer needs and serve them personalized offers suited to them on the basis of their own interests and activities on your website? Let’s see which approach is better in three A/B tests that compare personalized offers and bestsellers in various campaign types - home page, 0 search results and exit popup.
1. Check before you leave, A.K.A. exit popup
Industry: Home appliances
Use case: Our client tried to encourage customers to stay on the website a little longer.
Customers who have visited at least 2 products during their visit, and they were trying to leave the website have seen exit popups. It consisted of 4 products. Of course we have prepared two versions of the recommendations.
AB test assumptions:
- Users were divided into 2 equal groups:
- A got bestsellers,
- B got AI personalized recommendations. - Tests were run for 30 days.
Results: Tests showed that personalized recommendations win in every indicator: CTR, conversion, revenue.
2. No products matching the selection, A.K.A. 0 search results
Industry: Electronics
Use case: Client decided to show product recommendations on the zero search results page. Product frames has been shown to customer, who used search on the website, but there were no results found. The frame consisted of 16 items presented in the carousel.
AB test assumptions:
- Users were divided into 2 equal groups:
- A got bestsellers,
- B got AI personalized recommendations. - Tests were run for 10 days.
Results: After the test the customer had no doubts about which campaign should stay on the website: personalized recommendations were better in CTR, conversion and revenue.
3. Recommended for you, A.K.A. Home page campaign
Industry: Electronics
Use case: Client decided to present on the main page product recommendations for his customers. Product frames which consisted of 30 items has been presented in the carousel and shown to customers on the home page.
AB test assumptions:
- Users were divided into 2 equal groups:
- A got bestsellers,
- B got AI personalized recommendations. - Tests were run for 16 days.
Results: After the test the customer had no doubts about which campaign was more effective. Personalized recommendations were better in CTR, conversion and revenue.
Bestsellers – To do or not to do? That is the question
Am I saying that you should completely stop using top products on your website? Well... no. Having such additional frames may interest some users, especially if top sellers feature a type of product they are currently looking for. This kind of social proof may convince them to make a purchase. But I definitely discourage you from implementing bestsellers instead of personalized recommendations.
The tests presented above show that it’s easier to attract customer attention with offers suited for them, rather than products that perform best overall.